Successful pilot means more opportunities for secure data analysis

© Getty Images
The National Institute for Public Health and the Environment (RIVM), Roseman Labs and Statistics Netherlands (CBS) have successfully applied a new technique to enable data to be combined in a more secure manner. Secure Multi-Party Computation (MPC) allows several parties to conduct joint analysis without having to share the underlying data with one another. This technique is already used in the healthcare sector, for example: hospitals can conduct joint research into patient data without sharing the medical records of individual patients.

Sharing data securely

It is becoming increasingly important for researchers to be able to combine data from different sources. However, a lot of data is too sensitive to be shared with other parties. So how can researchers exchange and use one another’s data securely without having to make any personal or confidential data public?

This is where ‘privacy-preserving techniques’ (PPT) come in. The techniques include the encryption or anonymisation of data, for instance. But there are also approaches that enable researchers to conduct analysis together without needing to access one another’s data. This enables researchers to analyse and combine data while safeguarding the privacy of the individuals involved.

Three institutions were involved, RIVM, Roseman Labs and CBS, each with their own specific requirements. They came together to tackle the technological challenge of how to work with data that cannot be made available to other parties directly. The result is a technology that works, and will enable a great deal of new statistical research.

In the public interest

Fannie Cobben is Director of Innovation in Socio-economic and Spatial Statistics at CBS, and is responsible for privacy-preserving techniques. She sees the pilot as confirmation of CBS’s strategy. ‘We’re committed to serving the interests of society and promoting secure methods of using and sharing data. Privacy-preserving technologies help on both fronts: they make it possible to do more with the data that is available and to collaborate with external researchers and international statistical offices, while also preserving existing privacy and security frameworks.’

‘CBS is playing a facilitating role. Wherever possible, we are supporting researchers who wish to apply various PPTs in specific projects, provided that CBS is able - and permitted - to participate in those research projects. There is a lot of interest, particularly in the health care sector. And there are international developments, too: Eurostat is working on a new platform, and the UN’s PET Lab – which CBS is also involved in – is exploring similar frameworks.’

Pilot study

In 2021, CBS was looking for a case study in order to examine the use of PPTs. Roseman Labs – which was a start-up at the time – was also looking for pilot projects to learn from real-world experience. This led to a case study involving an analysis of sugar and fat consumption by geographical area. The RIVM provided nutritional data, while CBS provided sales figures. Normally, it would not have been possible to combine these two data sets for reasons of confidentiality.

‘You wouldn’t set up a system like that for a short-term study,’ explains Jan de Bruin, innovation manager at the RIVM. ‘And it’s a fine example of how new, innovative businesses can be given an opportunity within the public sector.’

Between 2022 and early 2025, the three partners completed a ‘proof of concept’ using dummy data. The analysis itself was relatively straightforward. The real added value lies in the ability to apply these techniques reliably in the privacy-sensitive context of the public sector. This requires complex IT integration in order to harmonise the operational requirements of the organisations involved and, above all, to integrate safeguards into data-processing procedures so that privacy is maintained.

‘Working with organisations in the public sector is highly valuable for a new business,’ says Toon Segers of Roseman Labs. ‘For us, this project has laid the foundations for larger-scale roll-outs elsewhere. It’s actually the best thing that any innovative company can have: a partner that wants to put the new technology into practice and is prepared to assess all the key criteria. And it’s even better if there’s a benefit to society at the end of it all.’

The assessment of the pilot was positive: the privacy and security criteria were met, and the analysis of the encrypted data was technically feasible at the required scale. For Roseman Labs, these projects and others like it were a valuable learning experience, and earlier this year the company went live with a major project involving several German banks.

Looking to the future

The Netherlands is at the forefront of this new technology: in academic circles, through the Centre for Mathematics and Computer Science (CWI) and Eindhoven University of Technology, and in practice, thanks to pioneering projects like the one carried out in 2018 by CBS, TNO and Zuyderland Hospital in Limburg. European supervisory authorities, including the European Data Protection Board (EDPB), are monitoring developments with interest.

‘If we hadn’t partnered with CBS and RIVM, we wouldn’t have been prepared for the Champions League of data processing,’ says Segers. ‘The government embraced this technology at an early stage, and its utility continues to grow. The National Cyber Security Centre has been using it since 2021. Another good example is the City of Rotterdam, which won the Netherlands Privacy Award for its system for continuously monitoring equality of opportunity in the city.’

For CBS, this means continuing to cooperate and collaborate, continuing to learn, and steadily rolling out proven techniques. Privacy protection and transparency need not be contradictory concepts, and we are constantly exploring new avenues for societally relevant research.

Cobben concludes: ‘Secure multiparty computation has certainly shown its value. The next step for us now is to look at how we can set up a production-ready environment. Wherever implementation is already possible, we will go ahead. And where further research is needed, we will continue to work with our partners.’